The proposed competing renewal builds upon the findings of the original grant showing that the frequency and density of self-references (I, me, my) are independently and prospectively related to the incidence of CHD in MRFIT, particularly CHD mortality. While these findings suggest that self-involvement may be a risk factor for CHD, the study was conducted with "high risk" males and the results may not generalize to males with normal risk factor levels. Therefore, proposed is replication using the Structured Interviews from another prospective study of CHD risk factors -- the Western Collaborative Group Study. We propose counting self-reference frequency and density from the baseline Structured Interviews for all 257 individuals who incurred CHD and a matched control sample of 514 who remained free of CHD. While self-reference frequency and density have been predictive of CHD, they remain a crude indicator of self-involvement. To further define how self-references appear and how they function in the Structured Interview, we have developed a more sophisticated analysis which we propose to apply to the original 577 MRFIT Structured Interviews. Specifically, this psycholinguistic analysis is designed to test hypotheses about specific aspects of self that may be related to CHD. The hypotheses will be tested with the 193 individuals who incurred CHD in MRFIT and their matched controls. In addition to the above findings for self-involvement, we have discovered a possible answer to the question of why Type A behavior did not predict CHD in MRFIT. Our results indicate that there are significant differences in the predictiveness of the interviewers' Type A assessments for CHD. Further, these differences are related to differences in the MRFIT interviewer style. Based upon a pilot study showing promising differences, we propose identifying specific interviewer speech characteristics that distinguish predictive from nonpredictive interviewers.